- Published: Tuesday, 09 February 2016 10:28
Sophia Bethany Coban
The SophiaBeads data provides a well-structured collection of data sets for testing reconstruction algorithms using different combinations of numbers of projections and exposure times, focusing mainly on comparisons of fixed total exposure type. The beads are simple and piecewise constant objects and thus appear well-suited for testing efficacy of total variation (TV) regularized reconstruction. For the purpose of studying if a sufficient number of projections for accurate TV-regularized reconstruction can be predicted by the image gradient sparsity (as simulation results published by JSJ suggested) we have faced a number of limitations listed below. These limitations caused a preliminary study submitted for the 2016 CT Meeting to be inconclusive on this question. Understanding a such potential connection may be of great practical use when planning experiments with time- or exposure-sensitive samples. We wish to acquire a new bead data set which extensively removes these limitations and hopefully allows us to reach stronger conclusions.
Sample & Safety
We are scanning 6 to 8 different sizes of beads in a single sample holder (so the object scanned at all times is a single plastic tube). The sample will consist of layers of these beads of different sizes, where we scan each layer by moving the sample up/down after a full scan.